数字制造系统现场总线暂态连接故障在线定位理论与方法研究

Network intermittent connection (IC) is a common but hard troubleshooting problem in networked manufacturing systems. It has direct impacts on the reliability of Industrial networks, and could turn network nodes offline and result in system-wide shutdown. However, due to lack of direct network reliability assessment tools, current network monitoring system can not evaluate the severity of the IC problem. Moreover, when network communications are significantly affected, there is no effective online fault diagnosis tool available to determine the correct location of the problem, which results in ineffective maintenance and the problem remains causing system-wide shutdowns. In this study, we propose a novel online IC fault diagnosis methodology for industrial networks. The major research focuses and expected scientific contributions are: 1) propose a new stochastic modeling method that integrate incomplete information of the event records and system states, and apply to IC problem modeling; 2) By intelligent querying the communication status of network nodes without significantly interfering the operation of the system, we propose a novel network reliability assessment method based on inference from the collected information, and apply to evaluate the severity of the IC problem; 3) Develop a new online fault localization method by integrating network topology information and network fault patterns. 4) Develop hardware in the loop simulation test-bed for digital manufacturing systems. The merits of this study is to lay a new foundation in methodologies and technologies for intelligent maintenance and safety monitoring on network manufacturing systems, and help the manufacturing enterprises improve their operations and quality of the products, which has a broad impact on national economy.

总线暂态连接故障在数字制造系统故障中占有比较大的比例，它是网络化数字制造系统常见却棘手的问题。现有的系统监控,在发现网络异常后,由于缺乏在线故障辨识及定位工具,无法找到故障的准确位置并予以排除,导致因暂态连接故障引起的系统停车现象重复出现。为此,本项目提出并研究数字制造系统总线暂态连接故障的在线定位理论和方法。主要研究内容及创新点:1)系统研究总线暂态连接故障的产生、发展规律及其描述方法,通过融合制造系统运行信息及非完整故障信息,建立总线暂态连接故障的时变随机模型;2)通过主动采集网络个别节点的通讯状态信息,提出暂态连接故障状态的评估方法;3)通过研究总线拓扑结构、故障特征、制造系统运行信息与故障位置的映射关系,提出暂态连接故障的在线精确定位原理和方法;4)构建数字制造系统半物理仿真系统。此项研究有望为数字制造系统的智能维护及安全运行提供新的理论、方法和技术支撑，具有重要的学术和应用价值。

本项目首先研究了数字制造系统现场总线暂态连接故障的溯源及发展规律；其次，研究了基于多端口的总线故障信息获取方法及技术，为故障诊断定位及总线健康评估打下技术基础；第三，研究了基于不完整信息混合事件流的暂态连接故障建模方法，以及基于部分可观测节点的数字制造系统总线暂态连接故障状态的评估方法，实现了总线系统的健康状态监测；最后，基于多端口数据同步采集，研究了基于复杂拓扑结构的数字制造系统暂态连接故障在线定位原理及方法。此外，还扩展研究了考虑总线暂态链接故障下的容错控制策略。本项目的主要成果有：发表SCI检索的国际期刊论文10篇（其中IEEE-TIE 2篇，影响因子7.05），EI检索的国际及国内会议论文9篇。另还有2篇期刊论文正在大修后的第二轮审评中（分别投稿至IEEE-TIE及IEEE-TVT）。授权国内发明专利 2 项；共培养博士研究生2人（已毕业1人，另1人将于2019年6月答辩），培养硕士研究生6人（已毕业4人，另1人将于2019年3月答辩）。成果已达到并超过了预期目标。